EmoDuck@sh.itjust.works to Memes@lemmy.ml · 2 年前Twitter users right nowsh.itjust.worksimagemessage-square360fedilinkarrow-up12.96Kcross-posted to: beyondcombustion@lemmy.beyondcombustion.net
arrow-up12.96KimageTwitter users right nowsh.itjust.worksEmoDuck@sh.itjust.works to Memes@lemmy.ml · 2 年前message-square360fedilinkcross-posted to: beyondcombustion@lemmy.beyondcombustion.net
minus-squareNoodlez@programming.devlinkfedilinkarrow-up4·2 年前Mm this could be a problem because server load is too unpredictable. I would actually say just randomize the list, so that it kinda does its own “load balancing” by incentivizing to pick whatever random top one it selected?
minus-squareSquare Singer@feddit.delinkfedilinkarrow-up6·edit-22 年前Yeah, whatever metric. Could also use a mix of number of users, some form of reputation measurement, uptime, etc. I mostly meant that the system should pick a “best server” and recommend that. Smarter people than me can come up with the best metric. But swamping the user with >100 servers to pick from is counterproductive.
Mm this could be a problem because server load is too unpredictable. I would actually say just randomize the list, so that it kinda does its own “load balancing” by incentivizing to pick whatever random top one it selected?
Yeah, whatever metric. Could also use a mix of number of users, some form of reputation measurement, uptime, etc.
I mostly meant that the system should pick a “best server” and recommend that. Smarter people than me can come up with the best metric.
But swamping the user with >100 servers to pick from is counterproductive.